RL in Scala
Gridworld
Pole balancing
Try it yourself
Watch it learn
Pacman
Slides
Source code
Grid
Controls
Step
Run
Pause
Rules
Agent can move Up, Down, Left or Right
Attempt to leave the grid → stay where you are, reward =
-1
Move in any direction from
A
→ jump to
A'
, reward =
10
Move in any direction from
B
→ jump to
B'
, reward =
5
All other cases, reward =
0
Q(s, a)
Policy